Demographics for all included participants.
| Demographics | ||||
| Summary | ||||
| N | Age (years) | Education (years) | Sex (M/F/O) | EHI |
|---|---|---|---|---|
| 555 | 29.42 (5.99) | 14.33 (2.42) | 301/247/7 | 38.29 (74.38) |
| Race | n |
|---|---|
| White | 408 |
| Black or African American | 50 |
| Asian | 48 |
| Multiple | 43 |
| American Indian or Alaska Native | 4 |
| Native Hawaiian or Other Pacific Islander | 1 |
| Other | 1 |
| Hispanic ethnicity | n |
|---|---|
| No | 494 |
| Yes | 61 |
Demographics for included participants, by handedness group (EHI bins).
| Handedness | N | Age (years) | Education (years) | Sex (M/F/O) | EHI |
|---|---|---|---|---|---|
| Left | 131 | 29.37 (6.34) | 14.16 (2.24) | 69/61/1 | -80.34 (19.8) |
| Mixed | 63 | 29.65 (5.9) | 14.76 (2.39) | 42/20/1 | -4.96 (26.52) |
| Right | 361 | 29.39 (5.89) | 14.32 (2.48) | 190/166/5 | 88.89 (15.36) |
| Left: (EHI <= -40) | Mixed: (-40 < EHI < 40) | Right: (EHI >= 40) | |||||
Within each handedness group, do we see the typical field x level
interaction? That is, do participants show a relative bias for global
shapes in the left visual field (LVF)?
Error bars show 95% CI.
Reaction time is modeled as a linear effect of field and level, using
data from every target-present trial with a “go” response:
lmer( rt ~ field*level*handedness + (1 | subject) )
| Field by level by handedness interaction (RT) | |||||||
| ANOVA: compare models with vs. without interaction term | |||||||
| npar | AIC | BIC | logLik | deviance | Chisq | Df | p.value |
|---|---|---|---|---|---|---|---|
| 9 | 809,804.09 | 809,885.087 | −404,893.045 | 809,786.09 | - | - | - |
| 10 | 809,804.25 | 809,894.246 | −404,892.125 | 809,784.25 | 1.84 | 1 | .17 |
| Field by level interaction (RT) | |||||
| Omnibus F-test | |||||
| term | df | sumsq | meansq | statistic | p.value |
|---|---|---|---|---|---|
| field | 1 | 532,813.683 | 532,813.683 | 7.665 | .006 |
| level | 1 | 6,371,638.985 | 6,371,638.985 | 91.665 | <.0001 |
| handedness | 1 | 3,373,026.257 | 3,373,026.257 | 48.526 | <.0001 |
| field:level | 1 | 2,270,639.42 | 2,270,639.42 | 32.667 | <.0001 |
| field:handedness | 1 | 967,375.584 | 967,375.584 | 13.917 | .0002 |
| level:handedness | 1 | 12,628.623 | 12,628.623 | 0.182 | .67 |
| field:level:handedness | 1 | 48,725.68 | 48,725.68 | 0.701 | .4 |
| Residuals | 59,843 | 4,159,669,538.499 | 69,509.709 | - | - |
| Field by level by handedness interaction (RT) | |||||||||
| Compare effect estimate to zero with emmeans() | |||||||||
| field_consec | level_consec | handedness_consec | estimate1 | SE | df2 | asymp.LCL3 | asymp.UCL3 | z.ratio | p.value4 |
|---|---|---|---|---|---|---|---|---|---|
| LVF - RVF | Local - Global | Right - Left | 10.322 | 7.61 | Inf | −4.594 | 25.238 | 1.356 | .17 |
| 1 A positive number means LVF global bias is stronger in right handers (as predicted by AAH) | |||||||||
| 2 Z-approximation | |||||||||
| 3 Confidence level: 95% | |||||||||
| 4 Two-sided | |||||||||
| LVF Global bias by handedness bin (RT) | |||||||||
| field_consec | level_consec | handedness | estimate1 | SE | df2 | asymp.LCL3 | asymp.UCL3 | z.ratio | p.value4 |
|---|---|---|---|---|---|---|---|---|---|
| LVF - RVF | Local - Global | Left | 16.929 | 6.505 | Inf | 4.179 | 29.679 | 2.602 | .009 |
| LVF - RVF | Local - Global | Mixed | 25.563 | 9.418 | Inf | 7.105 | 44.021 | 2.714 | .007 |
| LVF - RVF | Local - Global | Right | 27.251 | 3.935 | Inf | 19.538 | 34.965 | 6.925 | <.0001 |
| 1 A positive number means global bias (faster RT for global) | |||||||||
| 2 Z-approximation | |||||||||
| 3 Confidence level: 95% | |||||||||
| 4 Two-sided, uncorrected | |||||||||
In progress.
Model RT as a linear effect of field, level, and EHI (continuous):
rt_ehi_model <- lmer( rt ~ field*level*ehi + (1 | subject) )
## Use anova() on competing models to test 3-way interaction.
interaction_stats <-
function(model_with_interaction,
model_with_no_interaction) {
return(anova(model_with_interaction, model_with_no_interaction))
}
rt_model_no_interaction <- update(rt_model_ehi, . ~ . - field:level:ehi)
interaction_anova <- interaction_stats(rt_model_ehi, rt_model_no_interaction)
interaction_anova |>
as_tibble() |>
rename(p.value = `Pr(>Chisq)`) |>
format_p.value() |>
pretty_table() |>
tab_header(title = "Field by level by ehi interaction (RT)",
subtitle = "ANOVA: compare models with vs. without interaction term")
| Field by level by ehi interaction (RT) | |||||||
| ANOVA: compare models with vs. without interaction term | |||||||
| npar | AIC | BIC | logLik | deviance | Chisq | Df | p.value |
|---|---|---|---|---|---|---|---|
| 9 | 913,235.715 | 913,317.794 | −456,608.857 | 913,217.715 | - | - | - |
| 10 | 913,235.48 | 913,326.68 | −456,607.74 | 913,215.48 | 2.234 | 1 | .13 |
| Estimated global bias by field, for EHI of -100 | |||||
| contrast | estimate1 | SE | df | z.ratio | p.value |
|---|---|---|---|---|---|
| (LVF Local -100) - (LVF Global -100) | 32.144 | 4.728 | Inf | 6.799 | <.0001 |
| (RVF Local -100) - (RVF Global -100) | 16.347 | 4.728 | Inf | 3.458 | .003 |
| 1 Estimated global bias (ms) | |||||
| Estimated LVF Global Bias for EHI of -100 |
| LVF_global_bias |
|---|
| 15.797 |
| Estimated global bias by field, for EHI of +100 | |||||
| contrast | estimate1 | SE | df | z.ratio | p.value |
|---|---|---|---|---|---|
| LVF Local 100 - LVF Global 100 | 35.378 | 2.912 | Inf | 12.15 | <.0001 |
| RVF Local 100 - RVF Global 100 | 6.845 | 2.923 | Inf | 2.342 | .09 |
| 1 Estimated global bias (ms) | |||||
| Estimated LVF Global Bias for EHI of +100 |
| LVF_global_bias |
|---|
| 28.533 |
\[
28.533 - 15.797 = 12.736ms \\
12.736/200 = 0.064ms / EHI unit
\] Each unit change in EHI (-100:100) corresponds to a
0.064ms difference in LVF global bias. This is the
slope estimate given by the summary function:
summary(rt_ehi_model)
## Linear mixed model fit by REML ['lmerMod']
## Formula: rt ~ field:level:ehi + field:level + field:ehi + level:ehi +
## field + level + ehi + (1 | subject)
## Data: aah_for_rt_model
##
## REML criterion at convergence: 913219.1
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -4.865845 -0.594984 -0.166179 0.370206 7.221165
##
## Random effects:
## Groups Name Variance Std.Dev.
## subject (Intercept) 27265.5 165.123
## Residual 42379.5 205.863
## Number of obs: 67505, groups: subject, 555
##
## Fixed effects:
## Estimate Std. Error t value
## (Intercept) 676.7291428 8.0820990 83.73185
## fieldLVF -20.4882991 2.5020157 -8.18872
## levelLocal 11.5963014 2.5201333 4.60146
## ehi 0.1317520 0.0966802 1.36276
## fieldLVF:levelLocal 22.1648880 3.5619717 6.22265
## fieldLVF:ehi 0.0566721 0.0299355 1.89314
## levelLocal:ehi -0.0475090 0.0301604 -1.57521
## fieldLVF:levelLocal:ehi 0.0636820 0.0426037 1.49475
##
## Correlation of Fixed Effects:
## (Intr) fldLVF lvlLcl ehi flLVF:L flLVF: lvlLc:
## fieldLVF -0.156
## levelLocal -0.155 0.500
## ehi -0.458 0.071 0.070
## fldLVF:lvlL 0.109 -0.702 -0.707 -0.050
## fieldLVF:eh 0.071 -0.456 -0.227 -0.156 0.320
## levelLocl:h 0.070 -0.227 -0.454 -0.155 0.321 0.500
## fldLVF:lvL: -0.050 0.321 0.321 0.110 -0.456 -0.703 -0.707
In progress. Model accuracy as a binomial
effect of field, level, and EHI (continuous):
acc_ehi_model <- glmer( rt ~ field*level*ehi + (1 | subject), family = "binomial" )